EvE: Evolving Ensemble of Agents
An open-source framework from a university research group that wraps existing coding agents, such as Codex or OpenCode, into an evolutionary loop that co-evolves a population of candidate solutions alongside a population of agent guidance and skills.
EvE co-evolves two populations for a task where a result can be tested or judged, such as designing an algorithm, improving code, or solving a mathematical problem: a solver population of functional components within a repository, and an agent population whose guidance and skills are continuously refined. The user supplies a working environment, such as a codebase or a mathematical problem, the files a solution is allowed to edit, and scoring steps, such as shell scripts or agent-judge prompts, and EvE searches for strong solutions without a hand-crafted, task-specific workflow.
It orchestrates third-party coding agents rather than providing model access itself; each agent session authenticates with credentials the user configures for that provider, and the project states explicitly that it does not share credentials or bypass a provider's rate limits or other restrictions. Codex is the default backend, with OpenCode also supported.
It is written in Python, installs with uv, and is licensed under Apache-2.0; it comes from the Scaling Group at the National University of Singapore.